Ditch the cloud for local HA protocols


(misa) #1

The unacceptable problems–as characterized by ST itself–during the last few days clearly demonstrate the major flaw in ST’s all cloud architecture. There is no viable reason whatsoever for an entire home automation system to fail due to Internet connectivity. In fact, routing all HA functions through the cloud is a gigantic security and privacy hole that must to be addressed immediately.

Home automation protocols and devices are unstable as is. It’s baffling that ST has failed to provide the ability to isolate local automation functions to the hub and away from their cloud.

ST needs to revisit their basic design and offload local automation tasks to local devices. If ST’s current hubs cannot perform these tasks, I would be more than happy to purchase a new hub or an add-on device.


Why the cloud?
(Minollo) #2

Reliability and responsiveness are my main concerns with the ST’s cloud architecture; I can’t imagine ST being able to deal with the flood of events generated by all the HA devices of all users, plus all the smartapps scheduled events, plus the smartapps execution itself; it just seems unreasonable.

I do like the cloud based approach as a starting point: configure devices, develop apps, test them quickly; and, if you are fine with the inevitable drawbacks of that configuration (latency, reliability, and, if you really worry about that, security), fine - you can keep running things with a very inexpensive ST hub; but I don’t see philosophical reasons why I shouldn’t be able to say “ok, this is all good; but I really want to run everything local, even if I need to pay three times as much for a ‘smart hub’”. It’s fine (and inevitable) to still require Internet access for some operations; but I don’t see why turning off a switch if there is no movement in a room for 10 minutes should require dozens of roundtrips to the ST infrastructure…


(Chrisb) #3

See the thread on Hub 2.0. It’s exactly what you are looking for.


(Convinced ST will never be unbroken…) #4

While I would love to see a hub with local storage and the resources to execute programs, it can, and is being done in the cloud. I mentioned the other day that the first time I used my Google wallet card at 'In ‘N Out’, I received a notification on my phone before they could hand my card back to me. Impressive!

Now I’ve used this card at other places and not seen this kind of performance, but sure enough, on my next trip to 'In ‘N Out’, bingo! This is literally a 2 second transaction from their terminal, to their merchant services, to MasterCard, to Google, to my phone; and all this in a small town. So it can be done, and in my scenario, we are dealing with numbers of transactions that SmartThings numbers can’t even begin to compare with.

SmartThings is the first product I have purchased that originated from a kick starter campaign. Clearly they are a small outfit. Clearly they have growing pains. It is no secret they are renting cloud resources from AWS. The question is at what level, and whether their systems were properly designed in the first place to handle the load. Frankly, I think kick starter campaigns pretty much scream “WELCOME TO OUR BETA! We’ll get it all ironed out sooner or later, after we get your money”, and we should be prepared for that to some extent.

I have been patiently waiting for over 5 months… but 5 months is not a long time in software development circles, especially from a small shop. But SmartThings’ promise has now caught the eye of a big money investor, so the excuses of a small shop with little resources will soon evaporate. We’ve been told we’re getting a new hub next year. I am hopefull that new hub + more resources + updated system = a platform that works reliably. For now, the reality is that not a single week goes by without multiple issues (in fact, I had two errors this morning).


(Greg Fountain) #5

@misa has already been there with the same rant.


(misa) #6

I wouldn’t call the inability to lock my front door ranting.

I’m expending lots of time and money to put together a reliable HA system. Having to deal with problems caused by ST’s cloud is unnecessary and extremely annoying.

Am I in the minority with my concerns regarding privacy and security with this cloud architecture? This is a very serious issue which people in this form will go berserk over when (not if) ST’s cloud is compromised by hackers.


(misa) #7

I agree with you 100%.


(misa) #8

You are totally on point. We need to keep reminding ST that this unreliable cloud based architecture is unacceptable.


(Morgan) #9

@misa Did you not realize it was built on a cloud architecture when you spent all this money on your setup ??? You should really read the background before purchasing a system. Even if ST cloud was bulletproof, your local internet connectivity or other network connectivity could cause issues.

I hate not having local functions for my devices as well, that is why i got zwave remotes as secondary controllers and choose a lot of zwave devices over zigbee because of that. I’ve also set up almost all physical switches where I possible can.

I agree with you, but you really should have understood this possibility before you bought into the ecosystem of ST.


(Beckwith) #10

I would not consider a home automation solution that does not give me Internet access. I believe most feel the same way. As such, there are many aspects of a cloud architecture which are MORE secure than alternatives. Granted, SmartThings needs to enhance their security, two factor authentication being one of them, but this is not an issue with Cloud.

Assuming we need remote access, what approach are you suggesting SmartThings move to that is more secure?


(Chrisb) #11

Not really. They’re already working on Hub 2.0 that isn’t cloud based for basic level stuff. Look, I don’t disagree with what you’re saying. But ST has already acknowledged, more or less, that cloud based alone isn’t enough and an upgrade is already in the works.


(misa) #12

I understood that ST uses an architecture that needs highly reliable and scalable cloud service. I was also aware that the company plans to offload as much of the local processing as possible in the near future.

The problem is that ST repeatedly fails to provided acceptable cloud services. The company also seems incapable of anticipating looming problems. You can say that I’m ranting, however, having experienced these issues, my expectation from ST is to make the local hub their highest priority project.


((Possibly not the Matt you're looking for)) #13

Huzzah for 2-factor authentication. That’s a no-brainer, frankly, and I expect it is already on a whiteboard somewhere at ST.

Regarding cloud privacy and security, the notion that moving things out of the cloud necessarily makes things more secure is debatable.


(Paul) #14

(ActionTiles.com co-founder Terry @ActionTiles; GitHub: @cosmicpuppy) #15

You are definitely not alone in this wish. Hub 2.0 is recognition that local processing power must be an essential part of SmartThings in the future.

I have made the same “rant” since the Kickstarter Project was launched. Connectivity, hacking, speed / outages are legitimate concerns; but, frankly, “The Cloud” is no longer just a trendy concept: It is, for many products/companies, mission critical, very fast, robust, and secure.

That “fact” is making me ambivalent. Particularly reassuring on the “detached” argument, though, is that competitive products are building processing and intelligence into their designs (eg. ALYT), though they are far behind SmartThings in terms of market and experience.

I have faith that SmartThings sees reliability as core, and knows there are multi+layered options.


(Geko) #16

To be honest, I believe ST performs “on par”, if not better then anyone else in its peer category (i.e. cloud-based HA systems). Other players (Iris, Ninja, Wink) are also experiencing frequent service blackouts, latencies, etc. Which begs the question if the cloud-based HA is a viable solution at all. Yes, the cloud is all the rage and it’s getting more robust and ubiquitous. But… no matter what, it’s always going be less reliable and less secure than an autonomous solution.

Cloud is a perfect solution for distributed systems like social networks, electronic payment systems, etc. where the nodes are spread out all over the world. But how can the home automation system benefit from the cloud when most, if not all the nodes are located within a 1000 sf area? Surely, it makes tech support and trouble shooting easier because they can peek and poke into the customer’s data instantly, update backend software and such. But I believe that the cost of these benefits to the end user is too great, particularly if you intend to use your HA for home security.

ST have promised “some local functionality” in the 2.0 hub, but I believe the keyword here is “some”. Also, given the ST’s track record, I’m very skeptical it will see the light of day soon. And if I have to buy a new hub anyway, why not look at alternatives?


(Beckwith) #17

Correct me if I’m wrong, but Nest is skipping the hybrid (local/cloud) architecture and moving toward a total cloud architecture. I personally think this is the future and having a prominent player move in this direction supports that. I also feel the existing SmartThings cloud/hub is doable and SmartThings is still “cutting their teeth” on it so to speak. Regardless, SmartThings has already committed to more local control in hub 2.


(misa) #18

Yes, really, we do need to keep reminding them… ST hasn’t officially committed to any specific features/functionality/architecture for the next version of their hub, nor have they provided an estimated release date.

Based on what I’ve read here, there could be a Hub 2, at some point which might address some of the numerous usability, reliability, and security issues with the current architecture. I’ve been in the software development business for a long time, and this is just a common tactic from string people along–pushing vaporware to use and old phrase.

It’s time for ST to publish theirofficial product features/functionality list , along with the release date for the next hub.


(Chrisb) #19

“We will have a lot of details coming out in January.”


(Bruce) #20

Look for CES announcement, that would be the normal timing, and it is in “January”.